Diagnosis and Classification of Epilepsy Risk Levels from EEG Signals Using Fuzzy Aggregation Techniques

of four different types of fuzzy aggregation methods in classification of epilepsy risk levels from EEG Signal parameters. The fuzzy technique is the first level classifier which works on the EEG Signal extracted features (patterns) such as energy, variance, peaks, events, duration and covariance. These features are obtained from an epoch of 2 seconds in all sixteen channels. Each epoch is sampled at 200Hz and digitized. The risk level patterns obtained by fuzzy techniques have low value of quality value and performance index. The aggregation operator based optimizations such as Ordered Weighted Average (OWA), Max-min method; Max product method and Sum-product method are applied on the fuzzy outputs. Comparison of these optimizations is studied and analyzed for a group of ten known epilepsy patients. Training and testing are performed using 480 EEG signal feature sets of 2 seconds epoch obtained from routine clinical trials. To evaluate the optimization performance, we also employed free response receiver operating characteristics method with mean number of false positive. High quality value as 23.78 is achieved in OWA method and Max-Product method.